VIDEO
COMMERCIAL
CAPTIONED
Ego4D's narrations were written by annotators after the fact; HowTo100M's come from YouTube ASR. Here the person doing the task says what they are doing while they do it, in one unbroken take of 3 to 5 minutes with hands in shot. Cuts, silent footage, hands leaving frame and dubbed narration are sent back, and each contributor records one clip, so voices and settings are many across a global base. What you get is video plus a spoken, first-person account of intent, hesitation and correction, aligned to the frames because it was produced at the same moment: a caption source that makes action-segment labels much cheaper.
Volume
5,000+
Clips
Contributors
5,000+
Regions
100+ countries
Updated
What’s in the collection and how it’s delivered.
Data modality
Video
Annotated
Captioned (synchronous spoken narration)
Exclusivity
Non-exclusive rights
Capture tool
Phone camera (in app)
Use cases
Video-language alignment and dense captioning, speech-grounded action segmentation, VLA pretraining
Format
MP4 / MOV
Resolution
720p to 4K as recorded, 30 fps typical (MP4, MOV)
Total volume
5,000+ clips (approx. 2 TB)
Regions
100+ countries
Purity level
99%+ passed QA review
Where this data comes from and how it’s licensed.
Every file comes straight from the contributor who provided it and is rights-cleared before it’s listed, so your license traces back to the source.
Contributor
Provides the data and uploads it straight to Kled.
Kled
Clears the rights before anything is listed.
Your team
Licenses it under the terms below.
License terms
Non-exclusive commercial training license.
Consent
Each contributor opted in to this use when they uploaded.
Specs reflect the current collection and may change as new data is added. Final terms are set in your license agreement.
